Per-deal margin is the wrong place to look for the leak
Reading margin one deal at a time hides the pattern. The fifth-worst project quietly loses money for three quarters before anyone notices, because no leaderboard ranks the book worst-first.
A partner services firm at thirty people knows its three worst projects.
The fourth is a guess. The fifth quietly loses money for three quarters before anyone notices, because nobody looks at the whole book at once. The conversation about margin happens one deal at a time, in a project review, in a Tuesday standup, in a Slack thread that resolves with “we will keep an eye on it.” Nothing aggregates. The pattern stays invisible because there is no screen that shows the pattern.
The worst project is not a question about a single deal. It is a question about the whole portfolio. Most tools never roll up to that question.
Where per-deal margin breaks down
A deal has its own margin number. A project has its own actuals. A spreadsheet, built quarterly by finance, reconciles the two. The leaderboard, if there is one, is the spreadsheet, and the spreadsheet is three quarters behind real-time because that is how often someone has the patience to rebuild it.
The shared-delivery case is where the math gets worse. One project serves three deals. Naive per-deal math counts the cost three times, once against each deal. The deal-level numbers and the portfolio rollup stop agreeing. Reconciliation becomes a manual chase. The finance person who is good at the chase becomes the bottleneck on every margin question, and the firm learns to stop asking margin questions because the answer takes a week.
A second pattern: the deal that closed at a healthy margin and ran over by twenty percent. The per-deal view shows a green number. The variance against the plan is the real story. Nothing surfaces variance unless someone asks for it, and “ask for it” is the same friction that hides the leak in the first place.
What ranked worst-first looks like
The fix is a single screen.
The project list is the firm’s working surface anyway. Add the columns finance has always wanted to see. Billed to date. Recognized margin. Variance against the planned margin. Receivables aging strip alongside, because cash and margin are not the same conversation but they sit next to each other. A totals footer that sums the portfolio. Sortable, so the worst project floats to the top without anyone reorganizing the view.
The columns are role-gated. Cost behind cost-view. Margin behind margin-view. A delivery PM sees what a delivery PM should see. A finance lead sees what a finance lead should see. Sales sees the deal-side view, not the cost-side view.
The shared-delivery problem gets its own rule. A project that serves multiple deals has its cost split across the sharing deals in proportion to each deal’s contract value. One engine does the split. The same engine writes the deal-level margin and the portfolio rollup, so the two cannot disagree. A shared project’s cost is counted exactly once, fairly, by a method finance can defend without a whiteboard.
The discipline is not in the formula. The discipline is in the engine being shared. When the deal-level number and the portfolio total are derived from the same calculation, they reconcile by construction. There is no quarterly spreadsheet rebuild because there is no quarterly drift.
Why the whole-book view changes the conversation
A leaderboard ranked worst-first does something a per-deal view does not. It forces the firm to talk about the fifth-worst project, not the worst one. The worst project is already getting attention. The leaks the firm has not noticed are the ones in positions four through ten. They never come up in a project review because they look ordinary. They come up on a leaderboard because they are below the line.
The cultural shift is the point. Margin literacy stops being a finance person’s job. It becomes a screen the founder reads on a Monday morning, sorted, with variance visible, with shared costs split fairly. The conversation moves from “how did that one project go” to “which five projects are pulling the book down this quarter,” and the firm starts fixing the second question instead of relitigating the first.
What PartnerView ships
PartnerView’s project list carries optional billed, recognized-margin, and variance columns alongside the standard project fields. Sortable. Role-gated, so cost and margin views are visible only to roles permitted to see them. A receivables aging strip surfaces alongside the financial columns so cash and margin sit on the same screen. A totals footer aggregates the portfolio.
Shared-delivery work runs through one cost-attribution engine. A project that serves multiple deals splits its cost across the sharing deals by contract value share. The deal-level financial view and the portfolio rollup both read from that engine. The deal owner’s number and the finance lead’s number reconcile because they are the same number, derived once.
The leaderboard is a sort, not a separate report. The portfolio is the project list, ranked the way finance has always wanted to read it. The leaks below the worst-three line stop being invisible.